Model comparison
Kimi K3 vs o3
Kimi K3 is the stronger model overall, scoring 59.5 to 47.5 on the Noometry Index. o3 costs 1.7× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Last verified . 34 shared benchmarks.
Summary
- They share 34 benchmarks with published results for both. Kimi K3 scores higher in 8 categories and o3 in 2 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Kimi K3 leads 63.0 to 32.0.
- The biggest single-benchmark swing is ARC-AGI-2: 60.4% for Kimi K3 and 6.5% for o3.
- o3 is cheaper at $2 / $8 per million input/output tokens, against $3 / $15 for Kimi K3.
- Kimi K3 accepts more context: 1.05M tokens versus 200K.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| Kimi K3 | o3 | |
|---|---|---|
| Provider | Moonshot AI | OpenAI |
| Noometry Index | 59.5 | 47.5 |
| Released | 2026-07-16 | 2025-04-16 |
| Weights | Open | Proprietary |
| Context window | 1.05M | 200K |
| Max output | 1.05M | 100K |
| Input $ / M tokens | $3 | $2 |
| Output $ / M tokens | $15 | $8 |
| Results tracked | 53 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K3 leads
Kimi K3: 61.0 (#10), o3: 46.8 (#64)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| WeirdML | 82.6% | 52.4% |
| LMArena Coding | 1508 | 1408 |
| ALE-Bench | 1,524 | 933.55 |
| SWE-bench Verified | — | 62.3% |
| DeepSWE | 68.5% | — |
| FrontierCode | 44.2% | — |
| SWE-bench Verified (bash only) | — | 58.4% |
| Aider Polyglot | — | 81.3% |
| LMArena WebDev | 1654 | — |
| FrontierSWE | 25.9% | — |
| SciCode | 59.5% | — |
| GSO | — | 8.8% |
| CadEval | — | 74% |
Agentic & Tool Use Kimi K3 leads
Kimi K3: 41.8 (#20), o3: 34.5 (#44)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| APEX-Agents | 50.6% | — |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 30.8% |
| τ²-bench Banking | 37.1% | — |
| DeepResearch Bench | — | 45.2% |
| OSWorld | — | 23% |
| PostTrainBench | 32% | — |
| GBAEval | 48.3% | — |
| GDP.pdf | 19% | — |
| LMArena Search | — | 1144 |
| METR Time Horizons | — | 65.4% |
| Vending-Bench 2 | 5,165 | — |
Reasoning Kimi K3 leads
Kimi K3: 63.0 (#17), o3: 32.0 (#78)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| ARC-AGI-2 | 60.4% | 6.5% |
| SimpleBench | 60.7% | 53.1% |
| ARC-AGI-1 | 94.5% | 60.8% |
| CritPt | 23.4% | 1.4% |
| Chess Puzzles | 39% | 38% |
| LMArena Hard Prompts | 1496 | 1402 |
| Mystery Game Puzzles | 26% | 29% |
| DTBench | 91.2% | 84.8% |
| LMCA | 52.7% | 39.7% |
| Epoch Capabilities Index | 157.45 | 146.86 |
| ForecastBench | 61.1 | 62.5 |
| Kagi LLM Benchmark | — | 67.6% |
| NYT Connections (extended) | 93.6% | — |
| EnigmaEval | — | 13.1% |
| Surface Evolver Bench | 95% | — |
Math Kimi K3 leads
Kimi K3: 74.2 (#16), o3: 50.2 (#58)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 72.2% | 33.3% |
| OTIS Mock AIME 2024-2025 | 97.2% | 84.4% |
| LMArena Math | 1491 | 1426 |
| FrontierMath Tier 4 | 39% | — |
| MathArena Final-Answer Competitions | 87.8% | — |
| ProofBench | 87% | — |
| Omni-MATH | — | 71.4% |
| MATH Level 5 | — | 97.8% |
| FrontierMath (Feb 2025 set) | — | 18.7% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge Kimi K3 leads
Kimi K3: 63.2 (#21), o3: 54.6 (#52)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| GPQA Diamond | 93.1% | 81.8% |
| SimpleQA Verified | 50.6% | 49.4% |
| LMArena Expert | 1521 | 1402 |
| Humanity's Last Exam | — | 20.3% |
| MMLU-Pro | — | 85.9% |
| Confabulations | — | 14.4% |
| GPQA (HELM) | — | 75.3% |
Multimodal o3 leads
Kimi K3: 37.8 (#70), o3: 41.4 (#36)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| LMArena Vision | — | 1214 |
| GeoBench | — | 74% |
| VPCT | — | 52% |
| Blueprint-Bench 2 | 29.5% | — |
| Furniture Assembly | 34.2% | — |
Multilingual Kimi K3 leads
Kimi K3: 56.3 (#21), o3: 51.7 (#105)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| LMArena Non-English | 1466 | 1401 |
| LMArena Chinese | 1529 | 1437 |
| LMArena French | 1491 | 1430 |
| LMArena German | 1488 | 1420 |
| LMArena Japanese | 1487 | 1403 |
| LMArena Korean | 1458 | 1370 |
| LMArena Russian | 1482 | 1406 |
| LMArena Spanish | 1472 | 1395 |
Instruction Following Kimi K3 leads
Kimi K3: 77.7 (#14), o3: 72.8 (#127)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| LMArena Instruction Following | 1483 | 1368 |
| IFEval | — | 86.9% |
Long Context o3 leads
Kimi K3: 45.8 (#29), o3: 53.3 (#6)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| LMArena Longer Query | 1494 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench | — | 17.8% |
Writing & Preference Kimi K3 leads
Kimi K3: 76.6 (#4), o3: 63.5 (#64)
| Benchmark | Kimi K3 | o3 |
|---|---|---|
| LMArena Text | 1476 | 1410 |
| LMArena Creative Writing | 1454 | 1359 |
| EQ-Bench Creative Writing | 2082 | 1676 |
| LMArena Multi-Turn | 1488 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
| WildBench | — | 86.1% |
| EQ-Bench 4 | 1339 | — |
Frequently asked questions
Is Kimi K3 better than o3?
Kimi K3 is the stronger model overall, scoring 59.5 to 47.5 on the Noometry Index. o3 costs 1.7× less per token, which makes it the better buy when Kimi K3's lead doesn't matter for your workload.
Which is cheaper, Kimi K3 or o3?
o3 is cheaper. It lists at $2 per million input tokens and $8 per million output tokens; Kimi K3 lists at $3 and $15.
Is Kimi K3 or o3 better for coding?
Kimi K3 scores higher on coding benchmarks: 61.0 versus 46.8 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 200K.
How many benchmarks do Kimi K3 and o3 share?
34 benchmarks have published results for both models. Kimi K3 has 53 scored results on Noometry and o3 has 63.